## Predicting future FX support and resistance levels using mathematics

This is a discussion on Predicting future FX support and resistance levels using mathematics within the Forex forums, part of the Markets category; Last night I attached the predicted support and resistance levels of GBPUSD of the Volatility Response Model (VRM) for today. ...

 Dec 27, 2017, 10:56pm #31 Joined Dec 2017 VRM GBPUSD results Last night I attached the predicted support and resistance levels of GBPUSD of the Volatility Response Model (VRM) for today. Here are the results. Weekly levels in the top 30 minute chart, daily levels in the bottom 30 minute chart. Times are GMT-4 . EMA channel (4,7) included. Daily level 1.3372 was the day's low before rising to daily level 1.3432. There were some large gaps between levels. Predictions for tonight and tomorrow to follow shortly. Attached Thumbnails
 Dec 27, 2017, 10:58pm #32 Joined Dec 2017 GBPUSD predictions ending 5pm 28th Dec in New York I attach the VRM predictions for GBPUSD for tonight and tomorrow finishing 5 pm 28th December in New York. A description of the algorithm and format of the attached chart can be found in the first post of this thread. For anybody interested in predictions for other FX pairs just follow the link at the end of these two documents. Attached Thumbnails
 Dec 28, 2017, 8:34am #34 Joined Mar 2004 I am of the opinion that current maths is unable to predict the future movements of a stock with much precision. If it could there would be a lot of rich mathematicians. I even wrote to a prominent mathematician who has ben on the telly often but never received a reply. And why ? The problem is too complex. __________________ Love your own area. Long live NIMBYISM for quality of life. The Westcountry is being overrun with grotty new estates
Dec 28, 2017, 2:57pm   #35
Joined Dec 2017
Quote:
================================================== ==================

1.) The algo logic starts by treating time as the tick of a clock rather than a real variable. After all you can never reach the point in space given by velocity multiplied by time when time = PI hours (3.1415926...). The algo therefore uses as input data, daily high, low and close values. The algo uses highly non-linear maths to relate sequential daily data and needs at lot of historic data to find the function. When the function has been found then the future is just one step away. Just input today's high, low and close and out drops three levels for tomorrow. That is just the start. Now I can combine daily data and get high, low and close every two days. And repeat the process and get another three levels over the two day time scale. And repeat up to the 8 day time scale using high, low and close over 8 days. So that gives 24 levels altogether. Each triplet of high, low and close (Hn,Sn,Ln) you see presented each night relates to a timescale used for the calculation. The whole process is repeated every night using the daily high, low and close that came in. The whole process is repeated every weekend using weekly high, low and close data. This is where the lack of data posses a challenge. There is not that much weekly data available so the algo exploits the fractal nature of financial markets to overcome this restriction.

2.) The Sn levels presented each night and each week are sentiment levels at the nth time scale. Until the market breaks through the top most Sn then the market is still bearish. Similarly until the market breaks through the bottom of the Sn levels the market is still bullish at some time scale. So the highest and lowest Sn are important. I also believe that S1 is significant from experience. The Hn and Ln are the extremes of price movement that are predicted by the algo for each time scale. That is why I call the algo the Volatility Response Model. News comes in and the market becomes volatile. Quite often some levels are duplicated or close together within 2 pips so you can discard at least one of them. Usually there are less than 24 levels when you discard. To reduce the number of levels on your chart you could try adding the first 5 above and below the price action and add further levels as needed if the market becomes volatile.

3.) The algo is highly non-linear maths that finds a relationship between succesive hstoric high, low, close data. Once found it is set in stone. When the next future high, low, close data is added to the data set the algo absorbs it and predicts the next future time interval. The causal relationship established by the algo is that a mathematical relationship can be found between sequential high,low, close data. And this relationship remains the same between today's end of day results and tomorrow's results still to be determined. The predicted levels of the algo are the structure about which the price action will move tomorrow. Each night I present GBPUSD predictions and at the end of the market in New York I present the price action about the algo levels. The proof of the pudding is in the eating as the saying goes. Interestingly quite often algo levels today will be also levels tomorrow. That is a level found today at one time scale will become the same level tomorrow at another timescale.

 Dec 28, 2017, 3:11pm #36 Joined Mar 2004 I wonder if it possible to simulate quantum computing without all the high tech stuff ? So with programming maybe the average desktop could do this. Regretably I haven't the nuts and bolts skill. However there may be different alternatives to the present binary system which would allow it. Consider trinary. That is instead of 1s and 0s, you have +1s, zeroes and -1s. __________________ Love your own area. Long live NIMBYISM for quality of life. The Westcountry is being overrun with grotty new estates
Dec 28, 2017, 3:12pm   #37
Joined Dec 2017
Quote:
 Originally Posted by Pat494 I am of the opinion that current maths is unable to predict the future movements of a stock with much precision. If it could there would be a lot of rich mathematicians. I even wrote to a prominent mathematician who has ben on the telly often but never received a reply. And why ? The problem is too complex.
================================================== ======================
I agree that the future price action cannot be predicted all the time. But most of the time the market price action is trending. Overall sentiment that an economy is declining will cause the markets to ditch a currency. The VRM calculates long and short term trend channels to asses this type of trend. Intra-day markets still trend. I believe that these trends are between VRM levels I post each night. Then there is reaction to news such as interest rate changes or CPI. Then the market becomes very volatile and can move hundreds of pips. My experience to date has been that these large moves are between VRM levels. Because of limited market data the VRM can only calculate only 24 levels. When news is dramatic then the market price action can move to the very edge of the trend channels. The trend channels give you an idea of what can happen if news is extreme.

 Dec 28, 2017, 9:40pm #38 Joined Dec 2017 GBPUSD results for today 28th Dec Last night I attached the predicted support and resistance levels of GBPUSD of the Volatility Response Model (VRM) for today. Here are the results. Weekly levels in the top 30 minute chart, daily levels in the bottom 30 minute chart. Times are GMT-4 . EMA channel (4,7) included. Daily level 1.3395 was the day's low before rising to daily levels 1.3453/1.3455. After that there was a bounce off level 1.3432. There were some large gaps between levels. Predictions for tonight and tomorrow to follow. Attached Thumbnails
 Dec 28, 2017, 10:51pm #39 Joined Dec 2017 GBPUSD predictions to 5 pm December 29th New York time I attach the VRM predictions for GBPUSD for tonight and tomorrow finishing 5 pm 29th December in New York. A description of the algorithm and format of the attached chart can be found in the first post of this thread. For anybody interested in predictions for other FX pairs just follow the link at the end of these two documents. Attached Thumbnails
Dec 29, 2017, 12:22am   #40
Joined May 2012
Quote:
 Originally Posted by gka The proof of the pudding is in the eating as the saying goes.
In an earlier post I asked about your trade results to-date for obvious resasons because you understand the model more than anyone else. If you can't generate a positive expetancy over many runs then it is unlikley anyone else could.

Intuitively there are too many levels to trade from. I would be interested in your trade process to make it work. The problem is that a 24 hour window in FX is actually comprised of three sessions : Asia; European; and US. The volatility and price movements of each of the currency pairs are different during each of the sessions. I am unclear on how do you filter and then integrate that difference into your trade plan for each of the currency pair.
__________________
_______________

Those who know do not speak
Those who speak do not know
Since I have spoken, I do not know

Dec 29, 2017, 6:07am   #41

Joined Dec 2009
Quote:
 Originally Posted by gka Hi Everyone, My name is Graham and I am a mathematician here in Canada. There are lots of different technical analysis methods to determine future support and resistance levels. They can be subjective and depend upon experience. I have been wondering whether support and resistance levels can be determined mathematically and now have a working algorithm I call the Volatility Response Model (VRM). The VRM calculates future support and resistance levels either one day ahead or one week ahead, together with short and long term trend channels. These predictions are not extrapolations from the past. I attach a white paper describing the algorithm. I am looking for FX traders for feedback, discussions and opinions on the VRM predictions. To this end I will post predictions in this thread as markets progress. Here is the first set of predictions for GBPUSD in chart format starting 5 pm New York time tonight. The second attached paper describes the format of these predictions on the chart. They are essentially 24 weekly levels and 24 daily levels. The latter will last until 5 pm New York time tomorrow night. Tomorrow night after the close in New York I will post the GBPUSD market price action and the VRM levels on the same chart.
It wont work.

For any level in the market to be of significance, it needs to be known by many people - otherwise the market will not react there.

The levels that are known by the most people are the ones that are the most visually obvious.

Then there's levels driven by 'lines in the sand' drawn by governments, banks that have written barrier options etc.

So expecting math to come up with magic levels that only your formula knows about - is not realistic. Having 24 levels per day will give you enough for "confirmation bias" purposes that the market will move close to one of them anyway.

There is also the fact you are basing your research on the premise that markets can be predicted. I would first test that assumption before trying to find ways to make the prediction.
__________________
The proof of the pudding is whether there is actually any pudding in the first place.

Dec 29, 2017, 6:26am   #42

Joined Dec 2009
Quote:
 Originally Posted by gka I have spent a lot of time experimenting with the Volatility Response Model (VRM) using different timescales and EMA channels. At short timescales like 1 minute candlesticks there are more reversals that turn out to be retracements at larger timescales. At long timescales like 4 hours the news can have a big impact while you wait for the time period to end. The mid value (between bid and ask) 30 minute candlestick charts with a EMA (4,7) channel seem the best for me. I have not been using this timescale and EMA channel for long so do not have reliable statistics. I have had a run of 11 successful trades in a row. I am very interested in traders' experiences using the VRM predicted support and resistance levels and how they use the levels.
AKA Curve Fitting....
__________________
The proof of the pudding is whether there is actually any pudding in the first place.

 Dec 30, 2017, 1:31am #43 Joined Dec 2017 GBPUSD results for today 29th Dec Last night I attached the predicted support and resistance levels of GBPUSD of the Volatility Response Model (VRM) for today. Here are the results. Weekly levels in the top 30 minute chart, daily levels in the bottom 30 minute chart. Times are GMT-4 . EMA channel (4,7) included. Daily level 1.3433 was the day's low before rising to daily level 1.3529. The market could not pull the EMA channel through this level. There were retracements off daily and weekly levels back to the EMA channel. There were some large gaps between levels. I will refer to this chart and EURGBP and EURUSD charts shortly Attached Thumbnails
Dec 30, 2017, 2:22am   #44
Joined Dec 2017
Quote:
 Originally Posted by Brumby In an earlier post I asked about your trade results to-date for obvious resasons because you understand the model more than anyone else. If you can't generate a positive expetancy over many runs then it is unlikley anyone else could. Intuitively there are too many levels to trade from. I would be interested in your trade process to make it work. The problem is that a 24 hour window in FX is actually comprised of three sessions : Asia; European; and US. The volatility and price movements of each of the currency pairs are different during each of the sessions. I am unclear on how do you filter and then integrate that difference into your trade plan for each of the currency pair.
================================================== ===================

Your point that there are too many levels. I have seen trader's charts in other threads with even more levels. I have browsed through John Murphy's book on technical analysis so I know there are many ways to draw lines at different timescales carrying out many different types of analysis...Gann, Fibonacci, Wedge, Triangles.....and on. It's quite easy to generate hundreds of lines.

The VRM has a maximum 24 weekly levels and 24 daily levels. Weekly and daily levels are often duplicated exactly or within a couple of pips. Together with the suggestion that you only need 5 VRM levels above and below the price action means there are far fewer lines to deal with. Except when the market is extremely volatile (for instance a interest rate change) then you will not need all the levels. Tonight's VRM chart for GBPUSD needed 7 weekly levels and 8 daily levels.

I have attached the VRM weekly (top) and daily (bottom) charts for today showing you the 30 minute candlesticks and EMA channel for GBPUSD, EURGBP and EURUSD. These levels were on the website link found at the end of my descriptive documents that started this thread. These charts cover the full 24 hours you describe for markets in Asia, Europe and US. As you can see the EURUSD market moved from weekly level 1.1942 through to weekly level 1.2025 predicted by the VRM last weekend. A large gap good for trading. You will notice that at 13:00 (GMT-4) EURUSD reached 1.2025 at the same time that GBPUSD reached its highest closing 30 minute candlestick. This tells me that GBPUSD was following EURUSD at that time.

In between the USD sell off the EURGBP did not know what to do and never smoothly bounced off a VRM level. EURGBP was never an easy trade today because it was USD sell off day.

When it's a GBP sell off day then GBPUSD and EURGBP will follow each other and EURUSD will be all over the place and never bounce smoothly off VRM levels.

Because of arbitrage GBPUSD X EURGBP = EURUSD it is necessary to consider all three FX pairs at the same time and thus all three sets of VRM levels.

So the trading strategy from 5pm last night to end of today. GBPUSD and EURUSD bounce smoothly off VRM levels from the get go. EURGBP is nowhere near a VRM level at 5pm New York last night. So it's a sell USD day. How far will these markets rise? I conservatively set the T/P at 50 points. If I were not asleep at 5 am my time in the morning then I would go for even more points by following the EMA (7.4) channel and waiting for it to crossover as it bounces off a VRM level.

I have spent a lot of time looking at different timescales and I want to reduce stress from fear and greed and exhaustion. So 30 minute candlesticks do the trick for me. From your comments I understand you have far more trading experience than I.

As a matter of interest did you have 1.2025 as today's EURUSD high? EURGBP and EURUSD predictions for this week and today will be on the website link at the end of my descriptive documents until noon on Sunday.
Attached Thumbnails

Dec 30, 2017, 2:39am   #45
Joined Dec 2017
Quote:
 Originally Posted by DionysusToast It wont work. For any level in the market to be of significance, it needs to be known by many people - otherwise the market will not react there. The levels that are known by the most people are the ones that are the most visually obvious. Then there's levels driven by 'lines in the sand' drawn by governments, banks that have written barrier options etc. So expecting math to come up with magic levels that only your formula knows about - is not realistic. Having 24 levels per day will give you enough for "confirmation bias" purposes that the market will move close to one of them anyway. There is also the fact you are basing your research on the premise that markets can be predicted. I would first test that assumption before trying to find ways to make the prediction.
Thank you for your observations. Can markets be predicted? Why are there so many different approaches? Everybody is trying to find the solution! My approach is the same approach to describe the murmurations of starlings. If you look at just one starling in a flock then it's flight path is all over the place. Looks completely random. Yet together all the thousands of starlings make a cloud with an undulating surface. Does each starling know what every other starling knows?. Together do all the FX traders have some common quality. Banks such as the BoE can draw lines and Soros can break them.

I can only test my assumptions by doing the maths and by making the predictions and seeing what happens each day.

As I wrote in my last comment. Not all the 24 levels are needed because some are duplicates and some are just 2 pips close so can be discarded. Tonight the VRM GBPUSD chart at the end of the day needed 7 weekly levels and 8 daily levels.

Last edited by gka; Dec 30, 2017 at 2:46am.